期刊文献+

基于改进DenseNet的刺绣图像分类识别的研究 被引量:1

Research on Embroidery Image Classification and Recognition Based on Improved DenseNet
下载PDF
导出
摘要 针对中华传统刺绣工艺传承保护问题中的分类任务,传统的刺绣分类方法存在耗时长、精度低以及需要大量掌握专业知识的人力资源等问题;设计了一种基于改进DenseNet的刺绣图像分类识别方法;构建刺绣图像分类识别数据集;采用局部二值模式LBP、Canny算子边缘提取以及Gabor滤波等方式提取纹理特征,将不同特征图与原图合并为四至六通道图像数据集送入网络进行消融试验,扩充了数据集宽度;为稳定训练过程,加速损失收敛速度,提出引入SPP(spatial pyramid pooling)结构优化模型;为提高分类识别精度使用Leaky ReLU激活函数优化ReLU函数;实验结果表明基于改进DenseNet的刺绣图像分类识别方法可解决传统刺绣图像分类方法中存在的问题,改进后的刺绣图像分类模型与基准模型相比准确率提高了8.1%,高达97.39%。 Aiming at the categorize tasks in inheritance and protection of traditional Chinese embroidery technology,three are the problems of long time consuming,low accuracy,and massive human resources of traditional embroidery classification method,an embroidery image classification method based on improved DenseNet is proposed.The local binary pattern(LBP),Canny operator edge extraction and Gabor filtering are used to extract the texture feature and the original image,which are merged into four to six channels image data set,the data set is sent to the network to conduct the ablation test and expand the data set width.The spatial pyramid pooling(SPP)structural optimization model is proposed to accelerate the convergence rate of loss in the process of training.To improve the classification and recognition accuracy,the Leaky ReLU activation function is used to optimize the ReLU function.The simulation results show that the embroidery image classification and recognition method based on the improved DenseNet can solve the problems in the traditional embroidery image classification method.Compared with the benchmark model,the accuracy of the improved model is increased by 8.1%,which reaches up to 97.39%.
作者 刘羿漩 齐振岭 董苗苗 梁允泉 葛广英 LIU Yixuan;QI Zhenling;DONG Miaomiao;LIANG Yunquan;GE Guangying(School of Physical Sciences and Information Engineering,Liaocheng University Shandong Provincial Key Laboratory of Optical Communication Science and Technology,Liaocheng 252059,China;School of Physical Sciences and Information Engineering,Liaocheng University,Liaocheng 252059,China;School of Computer Science and Techology,Liaocheng University,Liaocheng 252059,China)
出处 《计算机测量与控制》 2023年第1期194-201,共8页 Computer Measurement &Control
基金 中央引导地方科技发展专项基金(YDZX2017370000283)。
关键词 刺绣图像分类识别 深度学习 卷积神经网络 稠密连接网络 金字塔池化 多通道融合 embroidery image classification and recognition deep learning convolutional neural network DenseNet SPPnet multi-channel fusion
  • 相关文献

参考文献9

二级参考文献117

共引文献107

同被引文献5

引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部